A major clinical problem in prostate cancer is that of tumor recurrence following initial apparently successful therapy. It is widely believed that recurrent tumors may arise from a small number of cancer stem-like cells that survive the initial therapeutic intervention and which have the capacity to regenerate the tumor. However, this idea has been difficult to test in vivo without manipulation of the cancer cells outside their native environment in the animal. Here we propose a lineage-tracing strategy to examine the competence of specific prostate epithelial cell types (castration-resistant Nkx3.1-expressing cells or CARNs and Bmi1+ cells) to regenerate tumors following androgen ablation in mice (Aim 1). We will further examine whether recurrent tumors driven by specific oncogenic mutations preferentially arise from particular cell populations of stem-like cells within a regressed tumor and the role of the androgen receptor (Aim 2). Finally, we will employ lineage ablation of specific cell types within a regressed tumor to assess the relative contributions of specific stem-like cells (CARNs and Bmi1+ cells) to prostate cancer relapse (Aim 3). In these studies will use novel approaches to define the significance of specific prostate cell types in tumor relapse following androgen ablation. This proposal if successful will have a major impact on our understanding of the cellular origins of recurrent prostate cancer and will facilitate effors aimed at successfully eradicating the tumor.